Traditional semiconductor integrated circuit technology is used to integrate various electronic circuits onto a common semiconductor substrate to form a system, or subsystem. However, the traditional approach to integrating circuits into a system has process, manufacturing and design limitations which make integrating some electronic circuitry onto a common semiconductor substrate impractical. A new integration technology, namely, system-in-package (SiP) technology, attempts to overcome the limitations of the traditional approach by interconnecting multiple discrete semiconductor systems on a common substrate and encapsulating the complete system in a common package. Generally, SiP enables the integration of a mix of technologies into one package that would otherwise be difficult and expensive using the traditional approach. For example, SiP technology has been successfully applied in mixed signal applications, such as RF/wireless applications and sensor applications, as well as in networking and computing applications, and other high speed digital applications.
As previously mentioned, the multiple discrete systems of an SiP are electrically coupled together to form a system and, as is well known in the art of digital electronics, many of the multiple systems communicate with one another by transmitting digital information in the form of electrical signals. Typically, even analog based systems included in the SiP have the analog signals converted into the digital domain. The electrical signals transmitted between the multiple systems represent a serial data stream where the data is represented as binary symbols having discrete levels of amplitude or phase, as well known. Multiple electrical signals are transmitted in parallel to transmit data of a data width, with each signal representing one bit of the width of data. Conventional signaling technologies between the multiple discrete systems in a SiP are generally based on standard off-die type signaling. However, it has been recognized that electrostatic discharge (ESD) requirements and load requirements can be relaxed since the signals are not being driven externally to the package of the SiP. In response, capacitively coupled signaling techniques have been employed between the multiple discrete systems of the SiP. Capacitively coupled signaling techniques provides various advantages over standard off-die signaling, such as, elimination of conventional circuits providing ESD protection, allowing devices having different voltage domains to be operated without level shifting, and low power consumption.
In a system using capacitively coupled signaling, the issue of DC balancing needs to be addressed. The requirement for DC balance in AC coupled signaling is necessary due to the nature of the bit patterns present in serial data streams. More specifically, long strings of ones or zeros can cause data recovery problems at the receiver due to the relatively constant voltage applied when strings of ones and zeros are transmitted charging and discharging the capacitive coupling between systems. As a result, the AC signal, which represents the transmitted data, can drift as the DC voltage level across the capacitive coupling changes. Additionally, the problem of DC balancing is exacerbated when the signaling is between devices operating in two different voltage domains, which may be the case in a SiP device. That is, the common-mode input at the receiver can vary enormously if the signal is not DC balanced, resulting in large common-mode variations at high bit-rates. As a result, data-recovery at the receiver is both difficult and complex.
One approach to the issue of DC balance has been to use Manchester encoding methods. Generally, in Manchester encoding binary digits are represented by a signal transition, and not the signal level, occurring within a bit boundary. That is, a “1” bit is typically represented by a rising edge of a signal (i.e., 0-to-1 transition) occurring during the bit period, and a “0” bit is represented by a falling edge of the signal (i.e., a 1-to-0 transition) occurring during the bit period. The encoding of bits in this manner may be alternatively viewed as a phase encoding where each bit is encoded by a positive 90 degree phase transition, or a negative 90 degree phase transition. Consequently, Manchester encoding is sometimes referred to as bi-phase encoding. Although Manchester encoding maintains DC balance, since any long strings of 1's or 0's results in a signal that oscillates between the high and low voltage values, it requires a bandwidth that is twice that of the bit-rate. That is, if the bit-rate is 400 Mbps, Manchester encoding requires a bandwidth of at least 800 Mbps. Thus, Manchester encoding sacrifices bandwidth by limiting bit-rates to no greater than one-half of the maximum available bandwidth.
Another conventional approach to the issue of DC balance is to use a “bit-stuffing” method. Generally, bit-stuffing involves “stuffing” extra transition bits into a bit-stream if a preset number of transitionless bits has been transmitted. The receiver follows the same protocol and removes the stuffed bit after the specified number of transitionless bits are received. For example, if the preset number is eight, and eight consecutive bits of a 1 or 0 are transmitted, the next bit, that is, the ninth bit, is inverted to cause a transition. Although bit-stuffing does address DC balance issues, encoding and decoding data at high speeds is non-trivial, and moreover, such a signaling scheme is inefficient since the “stuffed” bit does not carry any information.
Scrambling is another conventional approach that has been used to address the DC balance issue. In scrambling, the data is scrambled using a known pseudo-random sequence and then de-scrambled at the receiver using the same pseudo-random sequence. For example, an Boolean XOR function is often employed for scrambling and de-scrambling the data. However, even when using a pseudo-random sequence for scrambling, it is still possible for a string of 1's or 0's to occur. Thus, employing a scrambling method does not address the issue of DC balance entirely.
Therefore, there is a need for an alternative approach directed to addressing the DC balance issue arising from capacitively coupled signaling.